In meeting of the group of Altruismo Eficaz Madrid dedicated to the future of Artificial Sentience and its ethical considerations last 29/01/2019 we were proposing ideas about how to solve some of the hard problems of sentience, such as the problem of other minds (How can I know if someone feels? Do the machines feel?) Or establish what are the most successful hypotheses in relation to the sentience (The sentience emerges or is invoked? Is the sentience useful or inevitable?)
Three possible solutions were discussed:
- The first one is the proposed by Andrés Gómez Emilson from Qualia Research Institute, with a solution based on Mindmelding and Phenomenal Puzzles (Vídeo).
- The second is the project of simulation of hypotheses related to sentience proposed by Manu Herran.
- The third is restricted to the dualistic hypothesis (under an emergentist but not epiphenomic paradigm) that considers experiences to be useful and evolutionarily selected.
Next I will explain this third solution.
It is very common to assume or accept as valid an emergentist paradigm of sentience. This emergentist paradigm assumes that the sentence is created or emerges from a certain configuration of matter or more basic elements (information). In this paradigm there are two main possibilities:
- “Sentience is inevitable”: One of them is that the sentience emerges inevitably and this sentience has no effect on the material behavior, just happens. Like the noise of an airplane, which is not useful for flying, simply inevitable (metaphor of Yuval Noah Harari). Sentience would be an “epiphenomenon,” a “side effect.”
- “Sentience is useful”: the second variant of this emergentist hypothesis is that sentience emerges from matter and this sentience does have an effect on experiences, that is, it does have an effect on the material world. This second hypothesis assumes that matter, in certain cases, does not behave according to the laws of physics. If this second version of the emergentist hypothesis were true, we should be able to detect these unexpected behaviors of matter, produced as a result of the subjective states.
That is to say, assuming as certain the emergence of the sentience and the effect that in turn the sentience has on the matter, we could design an experiment to detect these unexpected behaviors of the matter, which are the result of the effect that subjective experiences produce.
What would this experiment be like? Although in reality there is probably much more suffering than enjoyment, and it seems that it is easier to generate pain than to generate pleasure, to design an experiment that would be ethical, we would provoke situations with pleasurable experiences and not with negative experiences. The experiment requires a system in which sentience could emerge and particularly, positive experiences. In that case, if really (positive) experiences were produced, and under the premises indicated above, we should be able to detect in the matter behaviors not expected according to the laws of physics. If a physical system had these unexpected behaviors, then we would deduce that it is sentient.
We can think of at least three ways of doing it:
- In a body. In a living biological body with a natural evolutionary origin, like any of us. If something gives us pleasure and this makes us perform an action, a sufficiently detailed observation of our nervous system should show that certain atoms or molecules react in function of a metaphysical event (experience) and not by physical causes.
- In a tissue. If we accept the idea that natural neuronal biological substrates are sufficient to generate sentience, we should be able to observe this phenomenon also in a neuronal tissue.
- In a computer. If we also accept an ontological antisubstratism, that is, if we consider that the material substrate is indifferent in relation to the sentience, we could even accept the logical substratum as sufficient for the generation of experiences. In that case, the experiment could be done in a computer, through a simulation of life. In this simulation, if the individuals felt, and this caused them to behave differently, we should be able to observe that the agents behave differently than they have been programmed, that is, breaking the rules with which they have been programmed.
In all three forms there is an important possible problem: it could be that in fact the matter or computer systems, assuming they are potentially sentient, are already behaving differently from the “laws of physics” and that what we call “laws” of physics “are in fact the” laws of physics and metaphysics “. That is, it could happen that we are systematically observing the effect that experiences have on the world of matter, but that we are mistakenly attributing this effect to a physical phenomenon (or categorized as unexplainable, but assumed as physical). One way in which this problem can arise, bringing the problem to a certain extreme situation, is panpsychism. That is, if panpsychism were true, this method would not be useful.
3 Comments